Nasiko-Labs/nasiko

The Open Runtime for AI Agents

What it solves

Nasiko addresses the fragmentation and lack of visibility when using multiple coding agents (such as Claude Code, Codex, and Cursor). It solves the problem of "invisible spend" where different agents use different billing units and dashboards, making it impossible to track total costs, attribute spending to specific users or models, or centrally manage model routing and tool permissions.

How it works

Nasiko acts as an "OpenRuntime" that sits in the call path of AI agents. It uses non-invasive discovery to detect existing agent harnesses on a machine and provides an optional routing layer. By pointing agents to the Nasiko LLM Router instead of a vendor's API, Nasiko can decouple the agent from the model provider, allowing users to route traffic to any supported LLM using a single set of credentials managed server-side. It also implements an MCP Gateway to provide a unified, permission-filtered view of tools and an A2A (agent-to-agent) proxy to enforce flow limits and security policies.

Who it’s for

It is designed for developers and organizations running a fleet of coding agents who need centralized observability, cost attribution, and governance over which models and tools their agents can access.

Highlights

  • LLM Routing: Decouples agents from providers, allowing a harness like Claude Code to be powered by an OpenAI model without changing application code.
  • TokenOps: Provides detailed cost and token attribution per agent and model using OTel spans.
  • Governance: Includes flow guards to prevent infinite loops between agents and an MCP Gateway for secure, centralized tool access.
  • Non-Invasive Integration: Detects and integrates with common coding agents (Claude Code, Codex, OpenCode, Cursor) without requiring SDKs or wrappers.
  • Embedded OCI Registry: Includes a self-hosted, S3-backed registry for deploying agents without external dependencies.

Related

  • Project
    zavora-ai/adk-rust

    A production-ready Rust framework (43 crates, 500K+ downloads) for building AI agents that work with many LLM providers, featuring tools, workflows, graphs, teams, realtime voice, coding agents, governance, and HTTP serving. Currently at 2.3.0 published, with 3.0.0 as an unpublished release candidate adding breaking changes.

  • Project
    smallnest/pigo

    pigo is a Go‑based CLI AI coding assistant (a clone of the pi agent) that can read/write files, run shell commands, fetch web content, and interact with many LLM providers. It works in headless script mode or an interactive REPL, supports tool‑level allow/deny lists, project trust, prompt templates, skills, plugins, self‑update, and provides a Go SDK for embedding.

  • Project
    atomicstrata/atomicmemory

    AtomicMemory is an open‑source, inspection‑ready memory service for LLM agents. It offers correction‑aware updates, model‑agnostic embeddings, and both managed‑cloud and self‑hosted Docker deployments. The project ships a Rust Core server, a TypeScript SDK, a CLI (`am`), adapters for LangChain, LangGraph, Vercel AI, etc., and plugins for various agent hosts. Quick start is a one‑line installer that creates a cloud profile; a `--local` flag runs the Core container on your machine. Memory can be ingested and queried via the CLI or programmatically via the SDK, and agents can be linked through an MCP server. Licensed under Apache 2.0.

  • Project
    ahmadrosid/nakama

    Nakama is an AI agent platform that allows teams to manage, configure, and deploy multiple AI agents with unique identities and tool access through a unified dashboard.

  • Project
    Dicklesworthstone/pi_agent_rust

    A high-performance, native Rust CLI agent for AI-assisted coding that prioritizes low resource usage, secure tool execution, and predictable terminal interaction.